=== himcesjf_ is now known as him-cesjf [12:35] Hi Kernel Team, I'm trying to bisect a bug which is in ubuntu releases, but doesn't show up in mainline builds. Is there some way to get the git history back to a mainline build? I can't seem to find this history in ubuntu-bionic.git [12:38] xandey: you can use the 4.15 tag, and the Ubuntu-4.15.0-* tags [12:46] cassardo: ah, i was looking for v4.15.17. Also, I wasn't able to build commits which have labels like "UBUNTU: Ubuntu-4.15.0-4.5" but aren't actually tagged... is it possible this is some sort of rebase or toolchain change? [12:47] cassardo: some of them ask me to change config options when following https://wiki.ubuntu.com/Kernel/BuildYourOwnKernel [13:15] xandey: yes, very likely a rebase [13:15] during development, those kernels are based on rc releases [13:16] further down the stable pipeline, you will more likely see kernels all based on v4.15 [13:16] and fast-forward, all of those reachable from the master branch, at least after the GA kernel [13:17] but as we are so close to GA, the oldest one you will find is Ubuntu-4.15.0-9.10 [13:27] caascardo: right, I confirmed that I get the issue with 4.15.0-9.10 (building myself and downloading prebuilt from launchpad). However, I don't see the issue in any of the mainline built kernels 4.15.0-rc1, 4.15.17, 4.17.0-rc6. [13:30] cascardo: I guess i'm looking to see if there's any more narrowing I can do at this point. But perhaps this is as good as I can do here, and it's time to try and debug directly. [13:34] what's the issue? did you report a bug already? [13:34] yeah: https://bugs.launchpad.net/ubuntu/+source/linux/+bug/1768139 [13:34] Ubuntu bug 1768139 in linux (Ubuntu) "Kernel crash when booting laptop with hdmi monitor connected" [Low,Triaged] [13:35] it's a kernel crash when a dell laptop is connected to hdmi and boots up, but only with ubuntu kernel [13:35] i was trying to do the rebase for the bug report [13:35] sorry, *bisect* [13:36] although I didn't see that last comment yet [13:52] xandey: you don't happen to have a Thinkpad X1, do you? [13:57] nope, dell insprion 15 [13:58] cascardo: Dell Inspiron 7559 [14:17] ok, i915 change in 4.15.0-9 was related to X1 only [14:40] well, it's not that I narrowed it down to 4.15.0-9, it's only that that's as far back as I was able to build. [14:42] Oh, looking back at my notes, it looks like I also managed to test 4.15.0-1_4.15.0-1.2 which asked me to change some cofig options, but also displayed the issue. [14:45] I could try reverting dc0f16f9b5084e6be2b8c79f8c6cd499a3451791 which appears to be the only change between 4.15 and 4.15.0-1_4.15.0-1.2 in drivers/gpu/drm/i915 [14:47] but I don't see that as a difference between the 4.15.17 from linux-stable and 4.15.0-24 [14:50] anyway, I don't mean to take up your time with a low priority bug, I just thought perhaps I was missing something with the build setup. [15:09] looking at the stack trace and the diff, it's hard to think this would be specific to ubuntu kernels [15:10] did you try our mainline builds? [15:11] yes, I tried the ones from http://kernel.ubuntu.com/~kernel-ppa/mainline/?C=N;O=D I also built the mainline myself [15:11] maybe some sort of module packaging difference? [15:14] well, enabling some drm debug could help [15:15] could you direct me a bit there? [15:16] drm.debug=0xe on kernel command line? [15:17] something like that, let me look at the interesting bits of i915 and see which flag would trigger them [15:23] =0x4 [15:23] KMS seems to have most of it [15:35] okay, i'll try that === himcesjf_ is now known as him-cesjf [16:42] cascardo: I grabbed two traces of the log, I think it will take me a while to puzzle out them. I posted them to the bug. [17:30] xandey: thanks [17:38] cascardo: no problem